GetMeThere

Inspiration

I think every one of us has had a public transport trip go bad. Unexpected congestion, bad weather, track maintenance, all contribute to network disruptions. We want to minimise the impact of those disruptions on everyday people.

What it does

We build a tool that will visualise the real time data and help us identify trends and patterns.. These trends and patterns can then be used to make suggestions to passengers to take alternate routes, or wait till the disruption has dissipated.

How we built it

We processed the Darwin data and loaded it into a visualiser written using Node.js.

Challenges we ran into

The dataset was actually quite small given that there are 200 different delay reasons. It was challenging to find any significant patterns, particularly as some delay reasons were not even present in the data. The other challenge was that disruptions tend to be different in different seasons - winter is typically very disrupted due to snow and frost causing signal failures and other track issues, and we only had data for the summer months. The other challenge was that many of the codes were too general and couldn't be used to effectively make predictions.

Accomplishments that we're proud of

The visualiser looks amazing! We like the concept of using the prediction data to advise customers to use alternate routes - we feel that as well as making the customer experience more efficient and comfortable, we have also provided a powerful tool for train operators to help manage disruptions pre-emptively.

What we learned

Prediction is heavily reliant on the quality and richness of the data, the more we have, the more we can do.

What's next for GetMeThere

Adding in information around the severity of the delay so we can use that to predict the duration of a disruption. Adding in additional data sources to act as a trigger for prediction, for example event data, weather data and scheduled track work and maintenance data.